High Occupancy Toll (HOT) lanes that use dynamic pricing to manage congestion and generate revenue are increasingly popular. In this paper, we estimate the behavioral response of drivers to dynamic pricing in an HOT lane. The challenge in estimation lies in the simultaneity of price and demand: the structure of dynamic tolling ensures that prices increase as more drivers enter the HOT lane. Prior research has found that higher prices in HOT lanes increase usage. We find that after controlling for simultaneity HOT drivers instead respond to tolls in a manner consistent with economic theory. The average response to a 10 percent increase in the toll is a 1.6 percent reduction in usage. Drivers primarily value travel reliability over time savings, although there is heterogeneity in the relative values of time and reliability based on time of day and destination to or from work. The results highlight the importance of both controlling for simultaneity when estimating demand for dynamically priced toll roads and treating HOT lanes with dynamic prices as a differentiated product with bundled attributes.